An Investigation of Forwarding in the MPLS support for Differentiated Services.

Abstract

The changing nature of the Internet-based applications is imposing stricter demands on the performance of the Internet. As the Internet resources become more and more constrained, the Best Effort (BE) model is increasingly proving less capable of providing the required Quality of Service (QoS). One of the solutions recently proposed by IETF is the Differentiated Services (DiffServ) architecture, which can provide different levels of QoS to each class by aggregating traffic into different classes at the network edge, and by giving differential treatment for each class within the core of the network. DiffServ, however, performs within the limits of the resources along the shortest path, and hence its performance is a function of resource availability along that path. Another standard proposed by IETF is Multi Protocol Label Switching (MPLS), a fast switching based technique that offers new capabilities for IP based networks. It combines the control of IP routing with efficiency of layer 2 switching. Traffic engineering(TE), or the ability to map traffic flows onto an existing physical topology is an example of key application of MPLS. MPLS and DiffServ, though independently developed, are complimentary technologies in the pursuit of end to end QoS.

IETF RFC 3270 provides a guideline and requirements for MPLS support for Differentiated services. We study RFC 3270 in terms of what it does and does not specify about MPLS -DiffServ . We investigate the issues involved in implementation of forwarding component in MPLS-DiffServ and evaluate the implementation vis-à-vis functional requirements, performance, and its ability to deliver better QoS. We conclude that MPLS-Diffserv does provide IP services a greater control over the network while simultaneously being able to deliver Different service levels.